I’m not a huge fan of Linkedin or resumes, so here we are — thanks for dropping by :D
I’m very interested the problem called AI alignment because I think it’s super important to figure out how to reliably control exceedingly advanced AI systems (we’re trying to build systems vastly smarter than us as fast as we can, what could go wrong?). I spend most of my free time doing research and increasingly thinking about scalable, maybe commercial solutions in AI safety. Aside from the above, I like to widen my training distribution (aka trying new things) and ‘build a world model’ by learning about different fields and problems (’grokking’ planes of legibility).
Ex-varsity table tennis and golf, always down to go hit some balls at the range!
This website is always a work in progress 😉
me^
Patterns and mechanisms of contrastive activation engineering
Y. Hao, A. Panda, S. Shabalin, SAR. Ali
ICLR 2025 Workshops on Bidirectional Human AI Alignment, Human-AI Co-evolution, and Building Trust in LLM Applications
Interpreting Large Text-to-Image Diffusion Models with Dictionary Learning
S. Shabalin, D. Kharlapenko, A. Panda, SAR. Ali, Y. Hao, A. Conmy
CVPR 2025 Workshops on Mechanistic Interpretability for Vision and Responsible Generative AI
Language Models for Open-ended Wargames
G. Matlin, P. Mahajan, I. Song, Y. Hao, R. Bard, S. Topp, E. Montoya, M. Parwani, S. Shetty, M. Riedl
EMNLP 2025 Wordplay Workshop
TODO
<aside> 🔭 Always down to connect & chat! → yap (https://cal.com/yixiong) → opportunities and collabs (lmk if there’s anything I can offer!) → build together!
</aside>
